The Genera of Bambusoideae (Poaceae)

(1)

The Genera of Bambusoideae (Poaceae) of the American Continent:

Keys and Comments

Cleofe E. Calderon and Thomas R. Soderstrorn

SMITHSONIAN INSTITUTION PRESS City of Washington

1980

(2)

A B S T R A C T

Calderon, Cleofe E., and Thomas R. Soderstrom. T h e Genera of Bambusoideae (Poaceae) of the American Continent: Keys and Comments. Smithsonian Contri- butions to Botany, number 44, 27 pages, 1980.-The history of the grass subfamily Bambusoideae is reviewed and nomenclatural problems of subfamilial and tribal level are explored. Characters are presented to distinguish the subfamily from all other grasses and differentiating features of the two major groups of bamboos-herbaceous and woody-are included. Keys are given for the tribes and genera of herbaceous American bamboos and genera of woody American bamboos, the latter based principally on vegetative characters. A conspectus of the subfamily also appears, with a list of all 37 genera recognized in the Ameri- can continent, each with nomenclatural and taxonomic notes. Comments on the morphology of the bamboo plant and the systematic value of some characters, especially vegetative, are given in the introduction.

OFFICIAL PUBLICATION DATE is handstamped in a limited number of initial copies and is recorded in the Institution’s annual report, Smithsonian Year. SERIES COVER DESIGN: Leaf clearing from the Katsura tree Cercidiphyllum japonicum Siebold and Zuccarini.

Library of Congress Cataloging in Publication Data C a l d e r o n , C l e o f e E . E.

T h e genera of Bambusoideae (Poaceae) of the American Continent.

(Smithsonian contributions t o botany : ‘no. 44) Bibliography: p.

1. Bamboo-America-Classification. 2. Grasses-America-Classification. 3. Botany-America -Classification. I. Soderstrom, Thomas R., joint author. 11. Title. 111. Series: Smithsonian Institution.

QKLS2747 no. 44 [QK495.G74] 581’.08s [584’.93] 79-14750 Srnithsonian contributions to botany : no. 44.

(3)

Page

Introduction 1

History of the Bambusoideae

. . .

4

Nomenclature 9 Characterization of the Bambusoideae versus Other Grasses . . . 11

Characterization of Woody versus Herbaceous Bamboos . . . 12

Key to the Tribes of American Herbaceous Bamboos . . . 13

Key to the Genera of American Herbaceous Bamboos

...

13

Key to the Genera of American Bamboos

...

15

. . . . . .

Conspectus of the Grass Subfamily Bambusoideae in the Western Tribes of American Herbaceous Bamboos

...

16

Tribes of American Bamboos

...

18

Genera of American Bambusoideae

. . .

18

Literature Cited

. . .

23

Hemisphere

. . .

16

iii

(4)

Dedicated to Professor C.G.D. Nees von Esenbeck (1776-1858) , father of the Bambusoideae Bambuseae, quidquid dicant alii, propriam subfamiliam s. ordinem graminunt eflormant,

cum vicinis omnino connexam.

Ruprecht, 1839

(5)

of the American Continent:

Keys and Comments

Cleofe E . Calderon and Thomas R. Soderstrom

Introduction

Bamboos are a unique group of arborescent grasses that occur throughout the world, although they are generally associated with Eastern cultures, where they play such a significant role in the every- day life of the people. T h e New World shares equally in the bounty of bamboos, bu t they are less known. T h e rich forests of the New World provided

man with all his needs and bamboo was only one of many useful plants at his disposal. While bamboo is used by man wherever he comes into contact with it, only i n a few places of the New World has he used it to any great extent, such as the lowlands of Colombia and Ecuador, where the native Gua- duas are used in construction.

Many bamboos, though, are herbaceous in nature and, looking like palm seedlings or ferns, inhabit the dark rain forests. Many of these are diploids, unlike the larger tetraploid and hexaploid bamboos.

T h e arborescent and the herbaceous genera, however, are similar in basic morphological and anatomical features and their alliance has been dem- onstrated by many corroborative studies. M’e consider all of the genera to comprise a single sub- family, Bambusoideae.

C l e o f e E . Calderon and T h o m a s R. Soderstrom, Department of Botany, National Museum of Natural History, Smithsonian Institution, Washington, D.C. 20560.

1

T h e concept of the subfamily Bambusoideae, with woody and herbaceous members, has evolved over a long period of time as part of the developing classification of the grass family. While the concept of the subfamily, as we espouse it today, was first presented by Ne=s von Esenbeck in the early part of the last century, few followed his ideas and only now, with additional data to support it, do most agrostologists accept the concept.

Unlike most grasses, the woody bambusoid species (bamboos) and herbaceous barnbusoid (which we call herbaceous bamboos) have biological peculiari- ties that have hampered their study. Most bamboos, for example, are large and complex plants that re- main in a vegetative state €or many years, flowering only occasionally. Collectors generally shy away from bamboos because of their size and complexity and usually because they are rarely in flower. Her- baceous bamboos, on the other hand, flower con- tinuously or at least for several months during the year, but their flowers are generally hidden under leaves or are inconspicuous, causing the plants to look sterile.

Nevertheless, there are many more collections of herbaceous bamboos than of bamboos themselves, and most collections of bamboos are incomplete.

O u r earlier field experience was directed especially toward the herbaceous bamboos, on which a greater number of studies had already been made than on

(6)

2

the bamboos. As a result we can now assign all the genera of herbaceous to tribes, while the state of knowledge does not allow this for all the bamboos.

It was McClure’s opinion also, as evidenced in his final paper, published posthumously (1973), that the American genera of bamboos were still too im- perfectly known to be placed in tribes.

Arborescent forms occur in a variety of unre- lated monocotyledonous families, for example the Palmae, Pandanaceae, Strelitziaceae, and Agavaceae.

T h i s growth form is also known in some members of otherwise herbaceous families: Aloe of Liliaceae, Puya of Bromeliaceae, and some genera of Araceae.

I n the same category the bamboos are one of the remarkable examples of arborescent forms in a pre- dominantly herbaceous family, the Poaceae. All of these arborescent monocotyledons exhibit funda- mental differences in construction, both from dicotyledonous trees and among themselves. Bam- boos represent one of the few groups of arborescent monocotyledons that compete with dicotyledonous trees in size and branching habit. Some species of Bambusa in South America may reach over 30 m in height and 20 cm in diameter. Examples of this and even greater size are known in Asiatic members, such as Dendrocalamus giganteus, the largest of all.

Bamboos, like the majority of arborescent monocotyledons, lack vascular cambia. Some essential features of the growth-form and growth-limiting characteristics in some arborescent monocotyledons have been described by Tomlinson (1964). I n the example that he gave, branching is in general considered limited, in some cases absent, or more commonly the production of lateral branches occurs only i n the inflorescence. It has been said that the absence of a vascular cambium from the aerial axis restricts branching in monocotyledons, while dicotyledonous trees with vascular cambia are typically much branched. T h e bamboos are probably the exception to this rule, and they present us with many questions still unsolved.

Very little is known about branching in bamboos from a morphological and physiological point of view. T h e production of lateral vegetative branches does not correspond exactly with any of the cases cited by Tomlinson. I n the aerial axis of a bamboo, lateral branches are produced from axillary buds, usually one at each node, although multiple buds are produced at the node in Chusquea.

A bamboo culm does not usually produce

SMITHSONIAN CONTRIBUTIONS T O BOTANY

branches at the lower nodes but only at some dis- tance above the ground, in the manner of a dicotyledonous tree. T h e number of branches at a node in the middle of the plant (“midculm node”) may be one or (some species of Merostachys) as many as 150 or more, I n many cases the original number of branches is augmented in successive years by rebranching from buds at the base of the branches.

T h e adult pattern of this set of branches presents very useful diagnostic characters of generic validity.

For descriptive purposes we refer to this aggregation of branches as a “branch complement.”

T h e underground axis of a bamboo also branches and produces a strong rhizome system. There are basically two types, sympodial and monopodial.

McClure (1966) refers to these as pachymorph and leptomorph, respectively.

For practical descriptive purposes we use the term

“bud complement” for the set of buds at the midculm node, which represents an early stage of the bud development and is typical for the genus. We avoid reference, however, to the number of initial buds since it can be known only from develop mental studies, which are still lacking.

Bamboos possess a basic pattern of construction that consists of a strongly segmented axis with solid or hollow internodes and with a particular system of ramification. T h e axis passes through different stages during its development. There are basically two phases, each recognized by a combination of morphological features. A detailed analysis of the growth habit of bamboos is outside the scope of this paper, mainly because complete developmental studies are largely lacking. For this reason we are not going to describe the morphological and physiological changes of a bamboo seedling, the very early stages of the juvenile phase, or the many intricate phenomena that occur with the onset of flowering.

We will refer very briefly to some features of the two major phases of growth form. These are different stages of the vegetative state of a bamboo plant and occur in the majority of known genera. As Tomlinson (1964) pointed out, bamboos constitute a peculiar and distinctive group with regard to their growth form and do not conform to any of the arborescent types that he defined.

Stem elongation plays a significant role i n the development of the plant form and consequently in the dimorphism of leaves, During the first phase they produce from the rhizome long shoots with

(7)

a strong apical dominance, fast internode elongation, and almost full suppression of lateral appendages, T h e new stem produces only one leaf at each node, in the form of a sheathing organ that en- velops it completely. Th i s new stem, which is at full diameter as it emerges from the soil, grows quickly and in many species reaches full height within a relatively short time. When the new shoot has completed most of its growth, the second phase of the vegetative cycle begins with the development of the lateral members, branches and foliage, at the same time that the soft tissues of the stem harden.

Th i s phase may extend over a period of several years during which the branching and rebranching occur along with a moderate increase in height.

I n the majority of grasses, including herbaceous bamboos, the development of the plant does not follow this pattern, which is associated with the arborescent habit. I n general, grasses produce short shoots with a balance between apical dominance, moderate stem elongation, and a simultaneous production of lateral appendages, one leaf at a node in alternate fashion but very few or no lateral branches.

Culm leaves and bud complements are the outstanding morphological structures of the first phase of the bamboo growth form, while branches and foliage leaves are characteristic of the second. These structures offer numerous characters of great systematic importance, which merit special comment.

T h e extreme dimorphism in bamboo leaves has given rise to much confusion in leaf terminology.

Th e r e ar e two basic types: “culm leaves” and

“foliage leaves.” Culm leaves cover the new growing culm and since their function is to protect the growing shoot, the greater part of the leaf is sheath;

often, in fact, the blade is hardly developed. Be- cause the blade is the less developed part of the culm leaf, it has generally been disregarded and the whole structure called merely “sheath” in the literature. Kurz (1876), in his perceptive account of the growth form of bamboos, drew particular attention to the useful characters in this structure to distinguish species.

Types of culm leaves are characteristic of genera, while differences in foliage leaves or in the culm leaf itself serve to distinguish species. T his structure, which is of paramount importance in identifying bamboos, is often overlooked by collectors.

As opposed to culm leaves, foliage leaves exhibit

a greater development in the blade or lamina, the sheath becoming more or less proportionate to the size of the blade or to the number of leaves in a branch. T h e blade i n most Bambusoideae is broad, more or less lanceolate, with the base constricted an d jointed to the sheath by a little stalk, which often bears a pulvinus. T hi s stalk is like the petiole of dicotyledonous leaves and is often similar in a p pearance; for descriptive purposes we use the term

“petiole” without any interpretive connotation.

Foliage leaves are produced laterally on the branches i n most cases and are replaced periodi- cally. T h e number of leaves on a branch varies within genera and species. Sometimes the leaves overlap closely, forming a distinct and recognizable set of leaves i n the distal part of a branch. I n our descriptions we refer to this group of leaves as a

“leaf complement.”

At this point we wish to stress the significance of the occurrence of two growth forms i n bamboos, a phenomenon that does not occur elsewhere i n the grasses. Moreover, the specialized leaves that cover the new shoot and later the numerous vegetative branches produced from almost every node are features unique among the grasses and rare in monocotyledons. Characters from these structures have proved to be of great systematic value at the generic and specific levels. Unfortunately these are the very structures that are usually not collected.

Flowers are of equal systematic importance i n bamboos as they are i n other grasses and flowering plants, but because they are so seldom available we must generally resort to distinguishing bamboos by vegetative characters, of which, fortunately, there are many.

After a limited period of vegetative growth the reproductive phase starts with the production of inflorescences. I n most bamboos this takes place at intervals from few to many years, at which times it is generally gregarious. I n some others the frequency of flowering i n a population is almost yearly, as i n species of Swallenochloa and Myriocladus, both genera that inhabit high mountains. It is interesting to note that many bamboos cease vegetative growth for a long period prior to the flowering and often lose all foliage at the peak of this process, which may last for a few years or more. I n many cases the development of inflorescences is not limited to the distal parts of the branches or culm, but to every part of the plant where meristematic tissue

(8)

4 SMITHSONIAN CONTRIBUTIONS T O BOTANY

allows it, from the base of the plant up. Thus, the complexity of bamboo inflorescences increases, mak- ing the traditional terms used in descriptions of grasses, such as spike, raceme, and panicle, inappro- priate.

Generally in bamboos and, to a lesser extent, in herbaceous bamboos, inflorescences are complex systems of ramification throughout which bracts and prophylla are well developed. T h e basic units of this system are partial inflorescences, in themselves clusters of spikelets. I n some bamboos the production of partial inflorescences starts at the base of the spikelet by the activity of buds in the axils of the glumes, as in the “pseudospikelets” of Bambusa (McClure, 1966, 1973). Along very general lines the inflorescences of bamboos can be divided into two major groups, I n one the production of inflorescences is restricted, limited to a single group of spikelets. I n the other case it is unrestricted, with unlimited production of inflorescences brought about by rebranching from buds at the base of any one of the partial inflorescences. T h e complexity of an inflorescence is increased by the different basic types of partial inflorescences themselves. These may be of either limited (determinate) or unlimited (in- determinate) growth.

T h e use of a very simplified terminology is highly desirable until detailed morphological analyses are carried out. For general descriptive words, such as

“bract,” “partial inflorescences,” or the concept of

“complex inflorescences,” we refer to Troll (1964). I n his more modern typological terminology the term

“synflorescence” could be apDlied to the bamhoo flowering system, but this would require an elaborate morphological analysis. More details on this subject are given in a previous paper (Calder6n &

Soderstrom, 1973), and for general meanings of different bamboo structures we refer the reader to the glossaries i n two of McClure’s publications (1966, 1973).

I n the following keys to bamboos we have given particular emphasis to the vegetative characters that are generally available, and for the herbaceous genera we have used both floral and vegetative characters since these are commonly present. I n the keys to bamboos we have added floral characters as supplementary information. Following the keys we have listed the genera of Bambusoideae an d for each genus we have given the author, type-species, type-locality, and a few notes on distribution and

ecology.

Because the concept Bambusoideae as we understand it today has taken almost two centuries to develop, we have followed its history and tried to clarify its development and present this in the following chapter. Confusion has reigned not only in the systematics of this unusual group of grasses but in its nomenclature as well, and only by understanding its history could we identify those responsible for conceptualizing the subfamily and the major divisions within it.

Rupert Barneby of the New York Botanical Garden and to Dr. Dan Nicolson of the Smithsonian Insti- tution for advice on nomenclatural problems and for their careful review of the manuscript, We owe a continuous debt of gratitude to Dr. Jose Cuatre- casas of the Smithsonian for his illuminating dis- cussions of morphology, advice on nomenclature, and ever-ready help in all botanical problems, big and small.

ACKNOWLEDGMENTS.-we are grateful to Dr.

History of the Bambusoideae

I n 1763 Adanson, in his treatment of Gramineae for the families of flowering plants, divided the family into several natural groups on the basis of a combination of characters, both floral and vegetative. His major groups are maintained to this day, the Poae and Panica being two of the most impor- tant, presently called tribes Poeae and Paniceae.

T h e beautifully executed and printed grass flora of New Holland (Australia) by Robert Brown in 1810 presented early ideas on relationships within the family. Brown remarked that, while grasses were connected in a reticulate fashion, he was nonethe- less able to recognize three groups, delimited chiefly by modifications in the fruiting (spikelet) structure.

T h e group of genera that contained Poa was characteristic of temperate regions and the other, which contained P a n i c u m , was characteristic of hot and tropical regions. He was familiar with both groups, the former i n his native England and the latter in tAle warm regions of Australia. Brown also listed a third group that contained a mixture of genera, which he admitted was scarcely natural. I n 1814 he contributed a treatise on the botany of Terra Australis (Australia) for the report of the voyage during 1801-1803 of the Investigator, of which Matthew Flinders was commander and Brown, him-

(9)

self, the naturalist. I n this account, Brown clarified his earlier remarks on grasses and divided the family into two tribes, which he called the Paniceae and Poaceae. He gave a very good explanation of the spikelets in each group and how they differ. Many later authors have interpreted these tribes of Brown to represent what many regard today as the subfamilies Panicoideae and Pooideae. While Adanson first named these two groups and characterized them, Brown elaborated upon them.

T h e genius of Charles Kunth is apparent early in the career of this German botanist who, at the age of 27, published a highly significant paper on grasses (1815) while he was at Paris. “Having had the occasion to examine a large number of species of grasses, living [plants] as well as [specimens] in the rich herbaria of this capital [Paris],” he was con- vinced that the genera could be ordered into many natural groups. H e presented a list of 10 such groups, among them the familiar Paniceae (as

“Gramina Panicea”). His tenth, the bamboos (as

“Gramina Bambusacea”), to our knowledge marks the first appearance of this group in a system of grasses. Kunth’s 10 groups were repeated in a later treatment of the grass family in his Synopsis Plan- t a r u m of 1822.

During the same year there appeared in Sweden the A p h o r i s m i of Agardh, in which nine groups of grasses were listed, one of them bamboos. It is likely that Agardh’s division of the family was modeled on the ideas presented by Kunth in 1815, since the groups are almost identical.

Kunth (1815, 1822b) and Agardh (1822) were thus evidently the first authors to regard bamboos as a separate category of grasses. They included the few genera known at the time.

I n 1827 Link listed the plants in cultivation in the botanical garden in Berlin, of which he was curator. Then, as now, one of the widely cultivated bamboos was B a m b u s a glaucescens, an oriental species with small foliage and slender branches, which adapts easily to indoor culture. Link called the bamboo Ludolfia and placed it in the Tri- glossae, one of two sections that he erected for the bamboos. His other section was called Bracteiflorae and in this he included N a s t u s and B a m b u s a a r u n - dinacea, the type of the genus, which is commonly called the “thorny bamboo of India.” Link’s sections were admirably based on characters from the inflorescence and spikelet, but faulty observations re-

garding stamen number caused him to include species of the same genus in each of his sections.

It was not until 1829 and 1835 that the bamboos were to receive a much more detailed treatment by the gifted Christian Nees von Esenbeck, a close friend of Goethe and a renowned botanist interested in all branches of natural history, philosophy, and social ethics. His 1829 treatment of the grasses of Brazil is a masterpiece. Bamboos constitute his tenth group, which he further divided into two parts, including in one of them the herbaceous genus Streptochaeta.

T h e 1835 work of Nees dealt solely with the bamboos of Brazil, a group that must have intrigued him during his earlier study of the grasses of that country. He now divided the bamboos into three groups: Bambuseae (including B a m b u s a ) ; Arundi- nariae (including Arundinaria); and a third, without a name, that contained Streptochaeta. However, the latter was listed as Streptochaeteae the following year (1836) in his enumeration of grass genera in Lindley’s N a t u r a l History o j B o t a n y .

I n the above system Nees gave the genus B a m - busa a detailed treatment and divided it into two subgenera, B a m b u s a and G u a d u a . T h e former contained species native to the Old World and the latter species native to the New World. Many authors have since regarded them as independent genera, but Hackel (1903) and McClure (1973) are among those who have followed Nees.

I n our opinion Nees’ 1835 paper was not only the first taxonomic monograph to be offered for bamboos but also one that embodied the most natural system, as we interpret it today, B a m b u s a and A r u n d i n a r i a being among our woody bambusoid grasses (bamboos) and Streptochaeta an herbaceous bambusoid (Calderon 8c Soderstrom, 1973). Further, B a m b u s a belongs to those bamboos that have sympodial (pachymorph) rhizomes and pseudospikelets, and A r u n d i n a r i a belongs to those that have monopodial (leptomorph) rhizomes and spikelets of the usual grass type. It is significant that Nees recognized the relationship of an herbaceous grass like Streptochaeta to the woody bamboos and ac- corded the genus equal rank. This paper is a testa- ment to the perceptive mind of one of the greatest agrostologists of all times, yet it went unheeded and unnoticed for over a century after its publication.

Few authors after Nees accepted his placement of

(10)

SMITHSONIAN CONTRIBUTIONS T O BOTANY

herbaceous and woody members together. Most, in fact, followed the lead of the Russian botanist, Franz Ruprecht, who, in 1839, published a large monograph on the bamboos. I n his lengthy introduction, Ruprecht remarked that, whatever others might say, bamboos form a natural subfamily or order of grasses wholly connected to the other members. Ruprecht placed the 67 species of woody bamboos in two groups, following the system of Nees.

Ruprecht’s contribution was not to the system of classification of bamboos but rather to the extensive worldwide treatment that he gave them. H e devoted several pages to the morphology of the plant, dis- cussed geographical distribution, and gave descriptions and keys to all the species. I n its restriction to woody bamboos, this monograph had a greater influence on later workers than did that of Nees.

T h e excellent monograph of Colonel Munro (1868) followed that of Ruprecht in its exclusive treatment of woody genera. Distinguished both as soldier and botanist, Munro was an outstanding agrostologist, who was often consulted by his colleague, George Bentham, on grass problems. Munro formulated three groups to accommodate his bamboo genera. T h e first two were those of Nees, but he added a new one, the Bacciferae, to contain the genus Melocanna and others from Asia. He divided the first group (which he called Triglossae out of per- sonal preference) into three parts, the Arundinariae, Arthrostylideae, and Chusqueae, the latter two containing the New World genera with which he dealt.

Munro’s system rested upon the foundation laid by Nees half a century earlier for the genera of Brazil but expanded to cover bamboos of the world.

His system was thus more complete and introduced a whole new group, the Bacciferae, of eight genera, all Asiatic, All the groups and subgroups that Munro recognized have remained in systems to this day, albeit at different rank or with different name.

Like Nees, Munro had a ke,-n perception of natural relationships in bamboos.

I n 1881 George Bentham published his classic paper, “Notes on Gramineae.” His friend, William Munro, had recently died and Bentham lauded the colonel upon whom he had depended for ideas regarding natural relationships in the grass family.

I n this paper, Bentham listed four groups of bamboos, adopting the concepts of Nees and Munro, with slight modifications in name a nd rank.

T h e treatment of grasses that appeared in 1887

in Die Naturlichen Pflanzenfamilien by the Aus- trian agrostologist, Eduard Hackel, has had a tre- mendous influence u p to the present time. T h e major divisions of grasses that he enumerated were followed by most subsequent investigators. H e divided the bamboos into four groups, which were exactly the same as those of Bentham. Hackel’s expertise in grasses such as Andropogon and Festuca did not extend to the bamboos, especially not to the herbaceous ones, which he dispersed among several distantly related groups.

T h e most recent treatment of grasses in Die Naturlichen PfZanzenfamilien was prepared by Robert Pilger (1954), who omitted the woody genera of Bambusoideae. H e did, however, assemble the majority of herbaceous monoecious bambusoid genera in one subfamily, which he called Olyroi- deae, and established another just for the herbaceous bamboo, Anomochloa. I n this context it is interesting to note that Pilger recognized nine subfamilies, three of which are based on genera that we regard as bambusoid.

At the time of Hackel’s paper, Franchet (1887) described three new herbaceous genera of grasses from tropical IVest Africa: Atractocarpa, Guaduella, and Puelia, all of which he stated to be bambusoid.

T h e superficial similarity in spikelets of one of the genera to those of the American bamboo, Guadua, in fact prompted i t s diminutive name.

T h e most complete monograph on Indian bamboos was published in 1896 by James S. Gamble, who was director of the Imperial Forest School at Dehra Dun, India, in the latter part of the past century. Although the monograph is a highly valu- able treatise, it did not contribute anything new to the system of classification since Gamble em- ployed the same four groups as Bentham.

I n 1913 Edmond-Gustave Camus published a monograph on bamboos, which was little more than a compilation of the literature to that time, although i t did present a slightly modified system.

His major groups were those already recognized by N u n r o except for the inclusion of Franchet’s herbaceous genera and some alterations in group names and rank.

Aimee Camus,like her father, Edmond-Gustave, worked at Paris an d published many new genera and spccies of grasses, especially from the French colonies of the time. Her classification of bamboos (1935) differed little from her father’s except for

(11)

the rearrangement of some existing tribes and the addition of two new subdivisions to accommodate recently described genera of Madagascan bamboos.

Both the Camusian systems incorporated woody genera. Interestingly, the only herbaceous ones were, however, those of their compatriot, Adrien Franchet.

T h e characters used in systems of classification u p through the time of A. Camus were for the most part floral, little importance being attached to vegetative structures. Probably for this reason Nees included Streptochaeta with the bamboos; the hermaphrodite spikelet, three lodicules, six stamens, and three stigmas were similar enough to those of bamboos to justify inclusion with them.

A few other authors (Meisner, 1843; Steudel, 1855;

Doell, 1880) followed Nees and placed that genus, along with Streptogyna (from tropical America, Africa, and Asia), next to bamboos. Trinius (1820), however, aligned Streptogyna with Bromus, and it remained with that genus through recent systems, an indication of his strong influence.

Other herbaceous broad-leaved forest grasses like Leptaspis, Olyra, Pariana, and Pharus, although closely resembling the above genera, were seldom associated with them. Grisebach (1864) did recognize the relationship of Pariana to bamboos, b u t most other authors misinterpreted its spikelike inflorescence as evidence of a close affinity to members of the Hordeae. T h e unisexual spikelets and monoecious condition, common to all of these genera bu t not features of any woody bamboo, probably account for the fact that these grasses were all placed elsewhere in most systems.

’IYhile Kunth, in his 1815 paper, regarded Olyra as representing a distinct group of grasses related to the bamboos, he was soon thereafter influenced by his contemporary, Carl Trinius, who treated the genus with Panicum in his monograph on that and allied genera (1826). Superficially the anthecia of Panicum and Olyra are indeed remarkably similar and many astute botanists since the time of Trinius have accepted without question the opinion of this influential Russian agrostologist. I n fact, this disposition was endorsed by Hackel (1887) and adhered to by his many disciples to the present. Some other authors felt that the monoecious condition of these genera was an indication of a natural alliance between Olyra, Zea, and Zizania, the latter two genera belonging to widely unre-

lated groups by our present understanding. While not linked in early systems with bamboos, some of these monoecious genera were placed together since early times: Pharus and Leptaspis, for example, had been associated with Olyra by Kunth (1815), Agardh (1822), and Link (1827).

It is well to recall the early words of Adanson (1763), who remarked that botanists were preju- diced in favor of divisions based o n floral characters. H e considered this a pitfall and, in establishing his natural groups of grasses, felt that all parts of the plant should be considered together.

But his advice was not heeded until 1931, when Avdulov brought together data from many lines of investigation that had been developing in the grasses since the end of the nineteenth century.

Study of the anatomy of the grass leaf has shown that there are great structural differences between major groups, and anatomical characters are now used widely in grass systematics. Of the numerous anatomical papers that we have previously cited (Calderon & Soderstrom, 1973:36), that of Brandis (1907) is of particular interest here. I n his well- documented paper, Brandis referred to the similarities of the structure between the leaves of bamboos and those of several herbaceous grasses such as Olyra, Diandrolyra, and Pharus. A short time later, o n the basis of leaf anatomy, Krause (1909) divided the family into three major groups, of which one was the Bambusoideae. T h e findings of Brandis were corroborated by Michaud-Page (1947) and Metcalfe (1956), who pointed out that leaves of Streptochaeta, as well as some other herbaceous genera, had the same anatomical structure as those of bamboos.

T h e embryo, endosperm, and seedling have also provided characters of systematic value. Van Tieg- hem (1897) studied the embryo and seedling and was the first to point out their value in distinguishing major groups of genera. His work has been added to by the embryo studies of Yakovlev (1950), Kinges (1961), and Reeder (1962). I n the latter study, the author pointed out similarities between embryos of the “olyroid” genera and bamboos.

Characters of the endosperm have provided the basis of a system of classification developed by Hayek (1925).

T h e detailed morphological and anatomical studies that became fashionable toward the end of the last century, and have continued since, were

(12)

8 SMITHSONIAN CONTRIBUTIONS T O BOTANY

directed as well to the grass inflorescence and spikelet. T h e unusual spikelet of Streptochaeta has intrigued most serious agrostologists since the genus was described by Nees in 1829 and placed by him with the bamboos. Celakovsky (1889) published detailed morphological studies of the spikelet and interpreted the flower as the most primitive of any grass, an idea that has persisted to the present. T h e classic paper on the morphology of the grass flower by Goebel’s student Schuster (1909) corroborated the opinion of Celakovsky. I n his phylogenetic scheme for the family, Schuster regarded Streptochaeta as the type closest to the hypothetical ancestor of the family. Several other investigators, such as Arber (1929) and Michaud- Page (195 I), have tackled this well-studied genus and published still further scholarly accounts.

T h e appearance of significant papers by Avdulov and Prat in 1931 marks this year as a milestone in the development of a modern system of classification for the grasses. Nikolai Pavlovic Avdulov was a cytologist at the Institute of Applied Botany and New Cultures in Leningrad, of which N. I.

Vavilov was director. Henri Prat was a French anatomist, who had recently completed his thesis on a study of grass leaf epidermis.

Avdulov, working under the cytologist G. A.

Levitskii, made a survey of the karyotype of all the grasses he was able to obtain. He studied the chromosome number, size, and form, in addition to the seedling and leaf anatomy. Combining his extensive data with those of others, he was able to develop an original system of classification.

Prat (193 1) showed that the grass leaf epidermis also provided numerous characters of systematic value. He expanded upon the data presented in that paper and in 1936 offered a new system, in- corporating his own data with those of Avdulov and others. While Avdulov had presented little data on bamboos-due to lack of study material- Prat recognized the bambusoid as a major division of the family.

Avdulov had set a precedent for evolutionary considerations in the grasses and made them the basis of a new phylogenetic scheme. His colleague, Roman Yulievic Roshevits, an agrostologist at the Botanic Garden in Leningrad, aided Avdulov in identifying the grasses that the latter used in his studies. I n 1937 Roshevits published a large treatise on grasses of the world.

In his classification Roshevits followed Avdulov with some modifications and presented ideas on evolution in the family derived mainly from Schuster’s phylogenetic interpretation of Strepto- chaeta and other bamboo genera. I n 1946 he presented another scheme in which his subfamily Bambusoideae contained six tribes, five consisting of herbaceous and one of woody genera. Hubbard’s system (1934), in which many new tribes were described, had appeared between the two papers of Roshevits and in the second the author adopts Hubbard’s new tribal names. In the 1946 scheme of Bambusoideae, Roshevits did not give a description of the subfamily nor his reasons for the tribal arrangement, stating that he would do so in a later paper.

Jacques-Felix (1955) published a paper on primitive African grasses, which he brought together in his bambusoid series. T h e author stressed the leaf anatomy common to all the genera, which he allo- cated to four tribes that included both woody and herbaceous genera. This paper established the modern concept of the Bambusoideae, based on a combination of natural relationships.

A few years later Parodi (1961) formalized the nomenclatural status of the subfamily Bambu- soideae and in his diagnosis added further sub- stantiating characters from embryo structure, chromosome type and number, geographical distribution, and ecology. In this particular paper Parodi dealt only with the genera of grasses from Argentina so that his subfamily Bambusoideae cir- cumscribed only the tribes with Argentinian rep- resentatives. I n agreement with this concept, we are following Parodi’s system, extending it to cover, in the present paper, the whole of the American continent.

A few systems of Bambusoideae have, however, confined themselves to the woody genera. Stebbins

& Crampton (1961), in their system of grasses for

temperate North America, felt that the commin- gling of herbaceous and woody members would result in a heterogeneous subfamily. They there- fore recognized only Arundinaria as bambusoid and relegated to the Oryzoideae the herbaceous genera Olyra and Pharus. Grosser & Liese (1973) presented a classification of the woody bamboos, based on characters from culm anatomy of a number of Asiatic genera they studied.

T h e most eminent bamboo specialists of recent

(13)

times have been Richard E. Holttum, of England, and Floyd A. McClure, of the United States, both of whom have made significant contributions to our knowledge of woody bamboos. Perhaps because both had such extensive experience with bamboos in the field, the former in Singapore and Malaysia and the latter in China and tropical America, they stressed the value of vegetative characters in this group, whose members flower so seldom, without devaluating those of the inflorescence, some of which are unique in the grass family.

Holttum (1956) commented upon some morphological characters that previous authors had used in their systems of classification and divided the bamboos into four groups, using characters from the gynoecium He did not attach formal names to the groups nor did he bring herbaceous genera into his discussion.

McClure (1961) limited his detailed description of the subfamily Bambusoideae to his own group of expertise, the woody bamboos. He did, however, later agree with the concept of Bambusoideae offered by Parodi in 1961, to which he refers in a revision of bamboos of Santa Catarina, Brazil (McClure & Smith, 1967).

N. N. Tsvelev, of the Komarov Botanical Insti- tute, has published a number of papers on evolution in the grass family. Of particular interest to us is his system of classification proposed in 1968, in which he divided the family into two subfamilies, the Bambusoideae and Pooideae, each further divided into a number of tribes. Tsvelev recognized the unique set of characters shared by genera of the Bambusoideae, which collectively form a group of equivalent rank to that formed by all other members of the family. I n his revision of the grasses of the Soviet Union, Tsvelev (1976) em- ployed this new system.

We recognize the Bambusoideae as a subfamily of the Poaceae, with a set of characters that binds together a large number of herbaceous and woody genera. T o most agrostologists today the Bambu- soideae is one of six or seven subfamilies of grasses; to Tsvelev it is one of two.

Nomenclature

I n 1812 Palisot de Beauvois described 69 new genera and 640 new species of grasses and proposed many new binomials and new names. H e divided the grasses into families, tribes, cohorts, and sec-

tions, and while all of these appeared in his

“Tabula Methodica,” none of them is given formal description. Trinius, a short time later (1820), in his Fundamenta Agrostog-raphiae, continued to use the old Linnean system, and his genera of grasses are to be found under such categories as “Trian- dria Monogynia” and “Monoecia Hexandria Mon- ogynia.”

As the number of described genera of grasses increased, especially during the first half of the last century, the idea of tribes was developed in order to bring together genera with similar characters. Robert Brown, in fact, used the term “tribe”

for the Poaceae and Paniceae, which he characterized in 1814. I n the following year Kunth stated

that “there is not at all a natural distribution of genera into tribes as has been done with success in other plant families. However, the possibility of this distribution should not be denied, which is the principal object of this memoir.” Kunth listed 10 tribes, for which he used such Latin group names as “Gramina Panicea” and “Gramina Stipacea.” I n a later paper (1822b), he repeated his 10 groups of 1815, but called each a “Sectio”

and gave new endings, for example, “Sectio 1.

Paniceae and Sectio 2 . Stipaceae.” Link (1827) offered a more elaborate system than his predeces- sors and divided the grasses into sections, but these were not comparable to those of Kunth, for his sectional category was superior to those of family, division, and suborder, terms that he used interchangeably. T h u s the Paniceae was a suborder of a section and the Stipaceae a division of his section called “Paniculatae Uniflorae.”

It would serve no purpose to review other systems of the time except to point out two items that are already obvious: (1) agrostologists recog nized that the genera of grasses could be grouped together on similar characters and (2) there was no agreement on names for these groups. For example, in the case of the well-known genus Panicum, Adanson established a section called “Panica.”

Another section was his “Poae,” which included several genera he felt related to Poa. Robert Brown recognized the Paniceae, calling it a “tribe,” characterized it further, and added more genera. Kunth recognized the same group, as one of his sections, but called it “Gramina Panicea” in 1815 and later just “Paniceae” (1 82213). T o Link, the Paniceae was a suborder of a section.

(14)

10 SMITHSONIAN CONTRIBUTIONS T O BOTANY

Regardless of what authors called the group of genera that included Panicum, they agreed that it was one of the natural divisions of grasses. Today we call a group of related genera a “tribe” and for this group we use the tribe Paniceae.

T h e first author to recognize and establish that Panicum and allies represented a natural group was Adanson. By present definition we regard this group as tribe Paniceae, of which the authorship must be credited to Adanson. Article 19 of the Zn t e 1.n a t i o n a 1 Code of B o t a n ica 1 N o m e n c 1 a t u re (ZCBN), note 2, states that when an improper ter- mination has been published, it should be changed to accord with the present rules. We thus change

“Panica” to “Paniceae,” but still give Adanson credit for establishing the tribe.

In the subfamily of special concern to us, the Bambusoideae, there is little agreement i n the literature regarding authors of tribes such as Olyreae and Bambuseae, or even of the subfamily itself. For example, Butzin (1973) credited Munro (1868) as author of the subfamily, while McClure, in his various publications, gave credit to others.

I n our opinion credit should go to Nees von Esen- beck (1835), who first recognized the bamboos and allies as constituting a major natural division of the family.

I n order to decide who was responsible for each category, we have followed the development of the concept of the bamboos as a subfamily of grasses.

I n tracing the ideas from one author to another, we have attempted to follow the development of the concept of natural groups of bambusoid genera (which we call “tribes”) and the aggregation of smaller groups into a large one (which we call

“subfamily”).

I n presenting the history of the Bambusoideae above, we referred to these aggregates of genera as “groups,” to avoid using the actual terms of the authors, since there was no uniformity in these terms. For the tribe Bambuseae, we thus give credit to Kunth who, in 1815, recognized the bamboos as a natural group. Nees von Esenbeck, in 1835, went further and established a larger group -subdivided into three parts, two containing woody genera and one an herbaceous genus-similar to the system we recognize today at the subfamily level. Tl’e thus give credit to Nees for the recognition of the bamboos as a subfamily of grasses.

Later authors recognized even further groups of genera of bamboos. Munro’s monograph of 1868 covered the Bambusoideae in the sense of our subfamily, even though he called it a tribe and named it “Bambusaceae.” Munro divided his Bambusaceae into sections and subsections, two of which, Ar- throstylidiae and Chusqueae, were recognized as natural groups for the first time. Today these groups are regarded as tribes and we give credit to him as their author.

Article 35 of the I C B N states that “a new name published on or after 1 Jan 1953 without a clear indication of the rank of the taxon concerned is not validly published. For such names published before 1 Jan 1953 the choice made by the first author who assigned a definite rank is to be followed.”

As we have pointed out, there was no uniformity in the names of rank applied to grass taxa i n the older literature. Since different names were used interchangeably for the same intended rank, they cannot be accepted in a literal sense. For example, Bambusa and allied genera constituted a section in the systems of Kunth (1822b) and Link (1827).

Several years later, however, Link (1833) made them a family “Bambusaceae,” which he listed as a subdivision of the section. On the other hand, Meisner (1843) called this group a subtribe.

We must look to the first author who recognized the taxon, whatever rank name he gave it, and assign the rank name that applies to the group by our present understanding. We have followed this way of thinking in our interpretation of the nomenclature of bambusoid taxa prior to 1 Janu- ary 1953, when guidelines were lacking. Since that date the rules for valid publication, as set forth in the ZCBX, must be followed.

Sometimes the selection of the authorship of a subfamily was based solely on the use of the first correct spelling by modern rules, an interpretation with which we do not agree. For this reason, McClure (1966) cited Rehder (1945) as author of the subfamily Bambusoideae of the “Gramineae”

and Ascherson & Graebner (1902) as authors of the subfamily Bambusoideae of the “Gramina.” It was not until late in the last century that the term

“subfamily” was in fact used as a category for a major subdivision of the grass family. h’ees von Esenbeck’s concept of the bamboos as constituting a major natural division of the grass family thus

(15)

preceded both the term we use for that group today and the rules that govern its correct Latin spelling.

Characterization of the Bambusoideae versus Other Grasses

BAMsusoIDEAE.-Perennial herbaceous or woody and arborescent, but without secondary thickening, rhizomatous; culms when woody branched at the nodes, the branches in some species with thorns or spines; leaf blades usually flat, broad, lanceolate or linear-lanceolate, articulate with the sheath by a petiole that twists and moves the blade in different positions; blades with tessellate venation strongly or weakly manifest superficially; flowering not seasonal, in some cases (bamboo) in cycles of long intervals of several years or almost continuous through several months of the year; inflorescences of different types, usually composed of complex systems of partial inflorescences with a limited or unlimited and continuing ramification and production of new partial inflorescences of restricted or unrestricted type (pseudospikelets), usually with very well- developed bracts (subtending leaves and prophylla);

spikelets or pseudospikelets 1- to many-flowered, without glumes or with 2 to several “transitional glumes;” lemma (3-)5- to many-nerved, not awned or only seldom so, and the awn then not geniculate;

palea 2- to many-nerved, keeled or rounded dor- sally, exceptionally bifid; lodicules generally 3 (rarely 0 to 6 or many), usually large and with hairs of different types and well-developed vascu- larization; stamens 3 to 6 or many (rarely 2), sometimes partially fused or monadelphous; stigmas 2 or 3 (rarely 1); fruit of variable type, dry (caryopsis or achene) or fleshy (drupe or berry-like), sometimes very large and many times the size of the original gynoecium; hilum linear, almost as long as the fruit; embryo several times smaller than the fruit.

Seedling: Coleoptile short and not elevated from the caryopsis by an internode, first two to several leaves bladeless or with a reduced blade, the first expanded blade broad, ovate or lanceolate, horizontal in position.

Leaf Anatomy: Leaf blade (lamina) with a very conspicuous midrib containing a complex vascular system of several bundles in two rows, strongly developed sclerenchyma and ground tissue; meso- phyll with cells arranged in horizontal layers par-

allel to the epidermis, not radiate; chlorenchyma composed of arm cells and translucent fusoid cells at each side of the vascular bundle and in between layers of arm cells; bundle sheaths always double and well developed, the outer sheath with very few chloroplasts; transverse veinlets connecting the longitudinal vascular bundles present.

Leaf Epideymis: With short cells in pairs or sometimes in rows over the veins; silica bodies usually cross-shaped, saddle-shaped, of olyroid type, or of intermediate forms; microhairs nearly always present, bicellular with both cells of about the same length and with rounded apex, microhairs of 3 or 4 cells sometimes occurring; papillae common and abundant on the long cells and overarching the stomata; long cells with thick sinuous walls;

stomata usually with low dome-shaped or sometimes triangular subsidiary cells.

Physiology: C, pathway of CO, fixation.

Cytology: Chromosomes small, x = 12, with most bamboos tetraploid (2n=48) or hexaploid (2n=72), and herbaceous bamboos diploid or polyploid, with n numbers of 7 , 9, 10, 11, and 12 reported.

Ecology and Geographical Distribution: Plants growing in association with trees or shrubs, usually in warm temperate woodlands, tropical rain forest or, if herbaceous, in the shaded understory of warm forests, also along streams or, if in rather open areas, always in shade of taller vegetation; usually dependent on humidity, shade, and warm tem- peratures in the environment; abundant in the tropics and subtropics of the world, only some woody members reaching temperate-cold areas of both hemispheres.

GRASSES (excluding Bambusoideae). - Perennial or annual herbs, sometimes tall and robust but neither with lignified culms nor complex ramification at nodes of the culm, nor thorns, nor spines;

cespitose, stoloniferous or rhizomatous, but then not with strong complex systems; nodes with only one bud or none; plants without a special growth pattern, i.e., not producing differentiated elongated new culms, the leaves all alike and produced pro- gressively as the plant grows; leaf blades usually linear and if lanceolate petiolate, not deciduous;

outer ligule never present; flowering seasonal, with one flowering period each year, either in spring or summer, according to the group; inflorescence spici- form, racemiform, or paniculiform, usually lacking bracts and prophylla and not forming complex sys-

(16)

12 SMITHSONIAN CONTRIBUTIONS T O BOTANY

tems of ramification; pseudospikelets not formed;

spikelets 1- to many-flowered, usually with 2 glumes; lemma awnless or with an awn, this sometimes geniculate; palea usually 2-nerved and 2- keeled; lodicules usually 2, relatively small; stamens usually 3; stigmas 2. Fruit a caryopsis, exceptionally of a different type, but then not fleshy: hilum punc- tiform, ovate or linear, short or long; embryo small or large.

Seedling: Of different types according to the group, but not bambusoid.

Leaf Anatomy and Epidermis: Of different types but not bambusoid.

Physiology: C, and C, pathways of CO, fixation according to the group.

Cytology: Chromosomes small or large, x = 7 , 9, 10, and sometimes 12.

Ecology and Geographical Distribution: Forming dense covers in open, sunny places, being the main component of prairies, steppes, savannas, and mountain slopes; common elements of the herbaceous vegetation from dry or arid zones to humid ones, from cold, temperate to warm, tropical, reaching the extreme limits of vegetation i n the mountains or near the poles of both hemispheres;

relatively few species adapted to the dark condi- tions of the woodlands or tropical forests.

Characterization of Woody versus Herbaceous Bamboos

WOODY BAMBoos.-Plants with a woody, strongly segmented cauline axis, often arborescent and with a prominently developed rhizome system; plants with a special growth pattern, producing at the initial period unbranched, elongated new shoots covered by overlapping modified leaves (“culm leaves”), and at the nodes one to multiple buds arranged in different patterns, leading when developed to complex systems of ramification; leaves dimorphic, those covering the new culm (“culm leaves”) with a greater deyelopment in the sheath, the blade usually smaller and of variable shape and orientation, caducous or persistent, the petiole sometimes lacking, auricles and oral setae often greatly developed; foliage leaves usually with a lanceolate blade, articulate to the persistent sheath by a petinle and deciduous, and often with tessellate

venation strongly or weakly manifest, outer ligule commonly developed, auricles and oral setae usually well developed; flowering in cycles of several-year intervals, the individual usually mono- carpic; inflorescences of different types in complex systems of ramification, with production of partial inflorescences either limited (one) or unlimited as the result of continuous ramification and production of several partial inflorescences of restricted o r unrestricted type (pseudospikelets); partial inflorescences composed of 1 to many spikelets arranged in racemiform or paniculiform fashion; spikelets or pseudospikelets 1- to many-flowered, the flowers all hermaphrodite; fruit a caryopsis, achene, or fleshy.

Plants common in tropical or subtropical lowlands, some reaching the temperate or cold regions and altitudes up to 4500 m, usually associated with woody plants.

HERBACEOUS BAMBoos.-Herbaceous, very small and delicate to robust plants, but without lignified culms and with rhizomes only weakly to moderately developed, never so strongly as in the bamboos;

plants without a special pattern of development, not producing tall, unbranched new shoots; nodes with only one bud, not developing vegetative branches or if exceptionally so, then the branch solitary; culms never armed; leaves not dimorphic, with only foliage leaves formed; blades lanceolate or exceptionally linear, not deciduous or exceptionally so, and rarely with manifest tessellate venation;

outer ligule generally lacking; auricles lacking or weakly developed; oral setae lacking except in the tribe Parianeae; flowering almost continuous or for several months of the year, not i n cycles of several years; inflorescences of different types, sometimes in complex systems resulting from the continuous ramification and production of new partial inflorescences; pseudospikelets not produced; spikelets l-flowered, hermaphrodite or unisexual (the plants then monoecious) or many-flowered, the flowers then hermaphrodite or rarely polygamous, the fe- male distal to the male; fruit a caryopsis, rarely a n achene. Plants common in forests of the tropics and subtropics, not reaching the temperate or cold regions or generally elevations greater than 1000 m, inhabiting the shade of forest or in more open places such as savannas, campos, or stream banks, sheltering under taller vegetation.